[Skip to Content]
[Skip to Content Landing]
November 15, 2010

Tread Carefully With Stepwise Regression

Arch Surg. 2010;145(11):1039-1040. doi:10.1001/archsurg.2010.240

In this edition of the Archives, Evans et al1 examine whether prehospital intubation results in higher rates of ventilator-associated pneumonia (VAP). They use logistic regression to study the impact of their main exposure variable (prehospital intubation) on their primary outcome (VAP). To adjust for other potentially confounding variables, they add multiple other independent variables to their logistic regression model. Evans et al find that prehospital intubation does not confer an increased risk of VAP. This analysis is a nice example of using multiple regression to study the relationship of an exposure and an outcome after adjusting for potentially confounding variables.